/ML

Personalizing The DoorDash Retail Store Page Experience

tl;dr: "In this post, we show how we built a personalized shopping experience for our new business vertical stores, which include grocery, convenience, pets, and alcohol, among many others. Following a high-level overview of our recommendation framework, we home in on the modeling details, the challenges we have encountered along the way, and how we addressed those challenges."

featured in #479

Ship Shape

- Kerry Halupka Rowan Katekar tl;dr: How Canva does hand-drawn shape recognition in the browser using machine learning to convert user-drawn scribbles into vector graphics, keeping classification latency at the forefront of the user experience. "We wanted to make sure the experience was snappy but still accurate. Therefore, we decided to deploy the solution in the browser, which allows for real-time shape recognition and drawing assistance, providing a seamless and interactive user experience. Users can draw shapes and receive immediate feedback without experiencing delays associated with server-based processing."

featured in #474

Navigating The Chaos: Why You Don’t Need Another MLOps Tool

tl;dr: AI/ML development lacks systematic processes, leading to errors and biases in deployed models. The MLOps landscape is fragmented, and teams need to glue together a ton of bespoke and third-party tools to meet basic needs. We don’t think you should, so we're building Openlayer to condense and simplify AI evaluation.

featured in #469

Building In-Video Search

tl;dr: "Suppose it’s Christmas, and you want to create a great instagram piece out all the best scenes across Netflix films of people shouting “Merry Christmas”! Or suppose it’s Anya Taylor Joy’s birthday, and you want to create a highlight reel of all her most iconic and dramatic shots. Creating these involves sifting through hundreds of thousands of movies and TV shows to find the right line of dialogue or the appropriate visual elements (objects, scenes, emotions, actions, etc.). We have built an internal system that allows someone to perform in-video search across the entire Netflix video catalog, and we’d like to share our experience in building this system."

featured in #464

Hey, Computer, Make Me A Font

- Sergey Tselovalnikov tl;dr: “This is a story of my journey learning to build generative ML models from scratch and teaching a computer to create fonts in the process.” FontoGen is a generative ML model project that crafts type fonts based on user descriptions. The author delves into the complexities of text-to-SVG generation and the intricacies of maintaining stylistic uniformity across glyphs. Drawing inspiration from the IconShop paper, a sequence-to-sequence model was employed, using text embeddings from BERT and font embeddings from tokenized glyph shapes.

featured in #454

Is This A Date? Using ML To Identify Date Formats In File Names

tl;dr: “To make it easier for our users to organize and find their files, Dropbox has an automated feature called naming conventions. With this feature, users can set rules around how files should be named, and files uploaded to a specific folder will automatically be renamed to match the preferred convention. For example, files could be renamed to include a keyword or date… We developed a machine learning model that can accurately identify dates in a file name so that files can be renamed more effectively.”

featured in #452

How DoorDash Improves Holiday Predictions Via Cascade ML Approach

- Chad Akkoyun Zainab Danish tl;dr: DoorDash's engineering team tackled the challenge of accurately forecasting supply and demand during holidays, where traditional tree-based machine learning models like Random Forest and Gradient Boosting faced limitations. The article introduces the "cascade modeling approach" as a solution. This method extends the Gradient Boosting Machine model with a linear model to account for holiday impacts, enhancing forecast accuracy. The cascade approach involves calculating holiday multipliers, preprocessing data, and post-processing forecasts.

featured in #446

Scaling The Instagram Explore Recommendations System

- Vladislav Vorotilov Ilnur Shugaepov tl;dr: Instagram has introduced a multi-stage approach to ranking, including retrieval, first-stage ranking, second-stage ranking, and final re-ranking. The system leverages caching and pre-computation with a Two Towers neural network, making it more flexible and scalable. Techniques like Two Tower retrieval, user interactions history, and parameters tuning - including Bayesian optimization and offline tuning - are employed. The article emphasizes the clever use of caching and pre-computation allowing for heavier models in ranking stages, and concludes with a note on the ongoing complexity and future improvements.

featured in #439

Leveraging Real-Time User Actions To Personalize Etsy Ads

- Alaa Awad Denisa Roberts tl;dr: Etsy has introduced a unique, innovative approach to personalizing machine learning models based on encoding and learning from short-term sequences of user actions. This is achieved through a three-component deep learning module known as the adSformer Diversifiable Personalization Module (ADPM). The module is aimed at improving the relevance of sponsored listings to the user's intent and is applied to the clickthrough tate and post-click conversion rate prediction models.

featured in #434

RLHF: Reinforcement Learning From Human Feedback

- Chip Huyen tl;dr: How exactly does RLHF work? Why does it work?” Chip discusses the  answers to these questions. “RL has been notoriously difficult to work with, and therefore, mostly confined to gaming and simulated environments. Just five years ago, both RL and NLP were progressing pretty much orthogonally – different stacks, different techniques, and different experimentation setups. It’s impressive to see it work in a new domain at a massive scale.”

featured in #414